Environmental Drivers of Plant Diversity and Turnover on the Guiana Shield

Elizabeth [Libby] O'Brien (UG), Aleksandar Radosavljevic, Robert [Bort] Edwards, Vicki Funk
Using 30 years worth of collections from the Smithsonian‘s Biodiversity of the Guiana Shield program we ask:
● What are the patterns of angiosperm richness? ● Which environmental variables are correlated with this richness? Are these the same variables that explain species turnover?

WHAT IS “THE SHIELD”? VARIABLE REDUCTION AND ASSOCIATION WITH DIVERSITY AND TURNOVER
• The Guiana Shield is one of the least explored regions in the world spanning
>2,287,000 km2 across parts of Venezuela, Brazil, Colombia, Guyana, French
Guiana, and Surinam.

• It is characterized by generally high rainfall, gradients in elevation, varied

topography ranging from open alluvial savannah to exposed rocky ‘tepuis’ (sheer
sided mountains), and a range of associated soil types.

• The flora of the region is noted for it’s high diversity and species turnover.
TURNOVER
• Relationship between these environmental factors and patterns of species
distribution are unknown, and identifying them will help us to understand the
evolution of the flora as well as to construct environment-to-diversity models.
Such models are important in prioritizing biodiversity assessment and
conservation, especially in the many areas currently inaccessible and poorly
explored. 4 a.

DIVERSITY

Figure 1. The Guiana Shield showing species diversity for 20km gridcells. Figure 2. Schematic showing reduction of initial environmental variables into sets, and result of species diversity and species turnover analyses.

METHODS DISCUSSION ONGOING WORK

• 149,762 collection records were retrieved from the Biodiversity of the Guiana
Shield database. These were reduced to 67,380 fully georeferenced and
standardized records after spatial and taxonomic cleaning.
• Values for 55 environmental variables (Soil - Soilgrids250; climate – WorldClim;
topography - NOAA) were extracted for each collection point.
• Records were aggregated to 1559 20km gridcells using Biodiverse with species
diversity metrics and average variable values calculated for each.
• Highly correlated variables (R2 >0.07) were eliminated (27 remaining variables).
• SPECIES DIVERSITY: A Principal Coordinates Analysis (PCoA) was used to
determine the strength of each remaining variable in explaining diversity.
Permutations of the data (999) were run to assess the significance of each.
• SPECIES TURNOVER: A nested series of Generalized Dissimilarity Models
(GDMs) were constructed to identify which variables explain species turnover.
• With the exception of soil pH, the spatial patterns of alpha and beta diversity are best
explained by two non-overlapping sets of environmental predictors.
• The number of statistically significant variables associated with alpha (14) and beta (7)
diversity also differs. This may reflect multiple species-rich communities associated
with different environments within the shield.
• Despite these differences, in both cases variables associated with edaphic conditions
and topography are important predictors of spatial patterns, highlighting the
importance of looking beyond climatic conditions to explain species’ and communities
distributions.
• Geographic distance and collection effort (number of samples per gridcell) are also
highly significant terms in models of species turnover. Accounting for these will be
important in predicting diversity into unsampled areas.
• With environmental layers available for the entirety of the Shield, the final
GDM model (using 5 variables) will be used to project species diversity for
areas currently without plant collection records. This will help guide
future work in species discovery, biodiversity assessment, and
conservation.
• Explore how species turnover changes across gradients in variables.
• Construction of a phylogeny for species in the environmental dataset is
ongoing. This will be used to compare phylogenetic diversity to raw species
diversity and further identify areas of conservation importance, as well as
evolutionary patterns (eg test between regional phylogenetic conservatism
or dispersal).
• Acknowledgements: This REU was funded by the National Science Foundation, grant number 1560088.
Background photo: Pixabay Free Images Online